Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is used to treat a wide range of malignant and non-malignant diseases. Despite the overall improvement in outcomes with allo-HSCT, acute kidney injury (AKI) remains a frequent complication. It contributes to the morbidity and mortality associated with this procedure1 and, in fact, patients undergoing allo-HSCT are exposed to many acute and chronic nephrotoxic events.

Several studies describe that pre-existing kidney disease, dehydration/fluid shifts, sepsis, the use of calcineurin inhibitors (CNIs) and nephrotoxic drugs (antibiotics, contrast medium and chemotherapy) are associated with kidney injuries during the first transplantation month (D+30).1-3 Other complications indicating kidney dysfunction, such as hepatic sinusoidal obstructive syndrome (SOS), tumoral lysis and thrombotic microangiopathy (TAM), seem to be less frequent. In contrast, beyond D+30, age and other conditions can significantly contribute to AKI, such as microangiopathy, graft versus host disease (GVHD), drugs, cumulative nephrotoxic events and viral reactivations, such as BK (polyomavirus) and adenovirus.4,5

The literature addressing this topic has been focused on the cumulative incidence of AKI until D+100 following allo-HSCT, instead of analyzing the hemodialysis (HD) use incidence, which may result in higher morbidity and mortality than AKI.1,5 On the other hand, studies have shown that some drugs (such as CNI) and successive AKI events in patients following allo-HSCT may induce cumulative kidney toxicity in the long-term, which could evolve to kidney function loss and, consequently, require renal replacement therapy (RRT).1-6 Some studies have shown mortality rates of 55 to 100% in allo-HSCT patients requiring RRT.1,5

According to the RIFLE (Risk, Injury, Failure, Loss of function, End stage) criteria, the diagnosis and severity of AKIs are based on urinary volume output, serum creatinine and the duration of kidney dysfunction. The reported incidences of AKI and dialysis in the allo-HSCT setting could range from 36% to 73% and 4% to 15%, respectively.1-6 The AKI consequences can trigger positive feedback to patient deterioration caused by hypervolemia, decreased ability to fight infection and platelet dysfunction.1-5 A new kidney injury classification, defined as KDIGO (Kidney Disease: Improving Global Outcomes), has been recently established and is considered more sensitive to detect AKI7 than RIFLE. However, the KDIGO7 criteria have not been routinely assessed in allo-HSCT patients yet.

The literature addressing hemodialysis in allo-HSCT is, therefore, heterogeneous, often reporting only short-term incidences,1,2,4,5 and risk factors for this severe complication remain poorly understood. We designed a retrospective single-institution study in Brazil to investigate the one-year cumulative incidence and risk factors for hemodialysis in patients undergoing allo-HSCT. We also investigated how hemodialysis might impact on patient survival.

We included a total of 185 patients undergoing allo-HSCT and their baseline clinical characteristics are shown in Table 1. Among these patients, 34 had to undergo HD during the follow-up. The one-year cumulative incidence of hemodialysis was 17.6%. In the HD-group, the median and mean time from allo-HSCT to hemodialysis were 55 and 99 days, respectively (IQ 25-75: 87 days; 95%CI: 61.6 - 137.6 days), while the time to hemodialysis ranged from 7 days (minimum) to 451 days (maximum).

We summarize the main allo-HSCT clinical outcomes, such as survival, relapses, GVHD and transplant-related mortality (TRM), in Table 2.

In general, the differences observed in baseline variables between the HD and HD-free groups were: haploidentical donor type (14/41.2% versus 38/25.2%; p = 0.051) and male gender 24/70.6% versus 79/52.3%; p = 0.021). Although not statistically significant, the proportion of patients with active disease status before HSCT in the HD group was superior to the HD-free group (13/40.6% versus 37/28%; p = 0.165).

Table 3 shows the results from the univariate analysis of potential risk factors for hemodialysis. Overall, the observed frequencies of nephrotoxic drug exposure, according to each group (HD versus HD-free), were: glycopeptide (vancomycin/teicoplanin) antibiotics (13/43.3% versus 92/65.7%; p = 0.037), liposomal amphotericin B (8/25.8% versus 49/34.3%; p = 0.406), amikacin (9/31% versus 17/11.8%; p = 0.019), polymyxin B (8/26.7% versus 9/6.25%; p = 0.003), and; cidofovir (20/64.5% versus 5/3.4%; p < 0.001). Only 57 patients had available data regarding foscarnet administration following allo-HSCT and the following differences were observed between the HD (4/10; 40%) and HD-free (6/47; 12.8%; p = 0.062) groups. When analyzing the duration of each drug exposure, we found that the median calcineurin inhibitor (CNI) duration was shorter in the HD group than in the HD-free group (137 versus 519 days; p < 0.001) and that there was no statistically significant difference in glycopeptide exposure in the HD versus the HD-free groups (16 versus 10 days; p = 0.088).

After a median post-HSCT follow-up of 408 days, 29.7% (52/175) patients experienced at least one episode of sepsis, requiring the vasoactive (norepinephrine) drug due to hemodynamic instability, in which a higher proportion of septic shock was found in the HD group versus the HD-free group (26/83.9% versus 26/18.0%; p < 0.001). Addressing thrombotic microangiopathy (TAM) in 97 patients with available data, the differences found between the HD and HD-free groups were 13/43.3% versus 7/10.6% (p = 0.001). The acute GVHD grades II - IV rate (20/58.8% versus 55/36.4%; p = 0.020) and steroid exposure (26/86.6% versus 86/59%; p = 0.039) were also associated with the HD risk; the latter was also included because it could be implicated indirectly in the HD through an increased risk of severe infections. Still addressing the infection risk, the median time to neutrophil engraftment was higher in the HD group than in the HD-free group: 17 days/range: 11 to 30 versus 15 days/range 10 to 22 (p = 0.007). Moreover, in the HD group, most patients (20/58%) were neutropenic (neutrophil count < 1.0 × 109/L) at one-week before hemodialysis.

The multivariate analysis revealed that septic shock/norepinephrine (hazard ratio, HR: 3.3; 95% confidence interval, 95%CI: 1.2 - 8.9; p = 0.024) and the cidofovir drug (HR: 11.0; 95%CI: 4.6-26.0; p < 0.001) were significantly associated with an increased risk for renal replacement therapy (Table 4), all of which are also represented with the respective cumulative incidences in Figure 1. The Haploidentical HSCT (HR: 1.94; 95%CI: 0.81 - 4.65; p = 0.14) and Age (HR: 1.01; 95%CI: 0.99 - 1.03; p = 0.18) were marginally associated with a hemodialysis risk.

Figure 1.

Cumulative incidence of hemodialysis according to risk factors. norepinephrine (hazard ratio, HR: 3.3; 95% confidence interval, 95%CI: 1.2 - 8.9; p = 0.024), cidofovir drug (HR: 11.0; 95%CI: 4.6 - 26.0; p < 0.001), haploidentical HSCT (HR: 1.94; 95%CI: 0.81 - 4.65; p = 0.14).

(0.36MB).

Addressing the impact of survival, according to the HD requirement, Figure 2 shows a significantly increased death risk of 6.82-fold (95%CI: 4.06-11.47; p < 0.001) in the HD group. At one year, the estimated incidence of death in the HD-free group was 21.2% (95%CI:14.8 - 29.7%), while for the HD-group, it was 75.0% (95%CI: 63.2% - 85.4%).

Figure 2.

Cumulative incidence of death (hemodialysis group versus no hemodialysis).

(0.28MB).

To better understand why 34 of our patients needed hemodialysis, each case was analyzed, discussed and reviewed together with nephrologists. Table 5 shows the main factors responsible for kidney failure, as well as each factor's contribution intensity in each patient, and the reasons why hemodialysis was indicated (as registered in medical charts). One case (that would be #34) was not shown in the Table due to incomplete data and inconclusive findings. We found that septic shock and cidofovir use were the leading causes of kidney failure. Additionally, some patients experienced TAM and SOS and other drugs, such as amikacin, amphotericin B, polymyxin B, foscarnet and glycopeptide, may have played a role in kidney injury. No GVHD-induced kidney lesions, such as nephrotic syndrome or immune-mediated glomerular disease, were observed.

Furthermore, two different patterns of creatinine kinetics before hemodialysis were detected in patients in the HD group, as illustrated in Figures 3 and 4. Figure 3 shows a subset of patients with a relatively stable and normal creatinine, who experienced sudden/rapid increment in creatinine, usually related to an acute event, such as septic shock. In contrast, Figure 4 shows another patient subset, presenting with an already affected creatinine, who experienced floating creatinine kinetics with slower increment, probably related to cumulative or multiple events related to kidney injuries, such as drugs use, virus infection and microangiopathy.

Figure 3.

Subset of patients with sudden increment in creatinine before hemodialysis (first pattern).

(0.31MB).

Figure 4.

Subset of patients with slower increment in creatinine kinetics before hemodialysis (second pattern).

(0.28MB).

The need for hemodialysis may have a substantial impact on allo-HSCT mortality. The literature shows that acute kidney injury is frequent in allo-HSCT, leading to high hemodialysis rates.1-3 However, the severe renal failure rates range very widely: from 4 to 49%.2 The 17.6% one-year cumulative incidence of dialysis seen in this study would be within this range cited in the literature. However, we believe that this rate is still too high and an “acceptable” rate has not been determined yet.

This unfavorable data may result from a combination of risk factors. Inappropriate or excessive indications for cidofovir use and a suboptimal approach to preventing bacterial sepsis could be the main issues.11 Moreover, because this is a cohort of patients treated in a private hospital, one could speculate that we might have faced a relatively high proportion of patients who had access to more treatment modalities and courses. Moreover, active disease status (30.5%) and haploidentical HSCT may contribute to an overestimation of the incidence of septic shock and severe viral infections. In addition, a longer duration of neutropenia and a higher proportion of acute GVHD in the HD group may play a role in the risk for infectious diseases (viral/bacterial) and thrombotic microangiopathy (TAM), as well. Indeed, an abusive administration rate (62.9%) and a prolonged use of glycopeptides (for a median time of 13 days) could be contributing to the HD risk. We believe that severe forms of hemorrhagic cystitis might have affected the kidney function of some patients, as well.

Another subset of our patients could subsequently be experiencing cumulative nephrotoxic events, as shown in Figure 4. These were also seen in one study, in which patients with early renal dysfunction, defined as a creatinine clearance < 60 ml/min up to D+90 following allo-HSCT, had increased risk (odds ratio (OR): 10) of developing a chronic kidney disease,12 indeed being more susceptible to kidney failure when affected by other nephrotoxic events. Although we could not assess it in our cohort, new biomarkers have been reported as factors significantly associated with more severe kidney damage in the pediatric allo-HSCT, such as serum cystatin C (cysC) and urinary neutrophil gelatinase-associated lipocalin (NGAL).13

The septic shock was an important complication in our cohort and one of the leading causes of kidney failure, experienced by 29.7% of the whole patient cohort and by 83.9% in the hemodialysis subgroup. Several studies addressed the need for admission in the intensive care unit (ICU) following allo-HSCT, which is usually caused by respiratory failure and uncontrolled sepsis,14-19 while we found one study15 that specifically focused on sepsis during the first year. The reported rate of ICU admission is heterogeneous, ranging from 11.4%19 to 13%,17 while a large case series reported up to 35%16 after a one-year follow-up. Therefore, these studies reported that septic shock incidences also range widely, from 3.217 to 15.7%.15 Indeed, they reported the duration of neutropenia as ≥ 14 days15 and acute GVHD17 as a risk factor for sepsis15 and UCI admission.17 Furthermore, the need for hemodialysis in the ICU is high, possibly affecting up to 66% of those patients.18

Despite the observation that polymyxin B and amikacin were associated with HD in the univariate analysis, we must interpret this data carefully, as these drugs are frequently used to treat septic shock, which was also identified as a risk factor for dialysis. The kidney function of these patients deteriorated rapidly (Figure 3; 48 - 72 hours), which might be a relatively short time to favor the possibility of drug-induced, rather than shock-induced, kidney lesion.

Before evaluating the impact of cidofovir on kidney failure, we must consider that the classical indications for this drug are adenovirus/BK-polyomavirus infections in a subset of severely immunosuppressed patients, which are frequently associated with hemorrhagic cystitis, viral-induced nephritis and microangiopathy, all known as nephrotoxic conditions, as well.2,5,11,20-22 Moreover, considering that in the haploidentical subset of patients, the incidence of hemorrhagic cystitis could reach up to 62%21 and in this context, we may explain why the haploidentical was also found as a risk factor for HD. Indeed, we believe that the TAM would also be contributing to kidney damage in our study, although it was not statistically significant in the multivariate analysis. As the TAM may be driven by CNIs, in this analysis we observed a shorter duration of the CNI in the HD group, probably because these patients died earlier; moreover, some stopped taking the CNI after kidney failure diagnosis due to suspected microangiopathy.

Despite the measures taken to prevent cidofovir-induced kidney injuries, such as using probenecid, hydration and dosing adjustments, unfortunately, our study found that cidofovir was strongly associated with the need for hemodialysis. Unacceptably, the subset of patients treated with cidofovir experienced a very high rate, up to 80%, of kidney failure, which is different from the rate of 9.3% reported by a systematic review.20 Another study had not observed an impact of cidofovir on the survival of patients with hemorrhagic cystitis.22 These findings, taken together with the observed toxicity profile, should make clinicians aware of the necessity to use this drug carefully.

The observed one-year cumulative incidence of renal replacement therapy in allo-HSCT was 17.6%. Patients in hemodialysis had a very high mortality rate, of up to 80%, at one year following transplantation. Clinicians must be aware of cidofovir use and adjust the conditioning intensity, calcineurin inhibitors and other drugs and immunosuppression to prevent patients from being exposed to bacterial and viral infections, microangiopathy, shock and other nephrotoxic events. This is even more important because these interventions may have a greater impact in the haploidentical HSCT patient subset.

none.